CS PhD student & @KnightHennessy scholar @Stanford. Prev.: @UniofOxford with @rhodes_trust, @Harvard '23, @MPI_IS, @CSatETH, @IBM. Theory CS for Trustworthy AI
Joined April 2013
- Tweets334
- Following2.4K
- Followers900
- Likes6K
We’re very excited to be organizing a 1-day workshop on “Recent advances in Regularity Lemmas (with and without calibration)” at Stanford on Tuesday, November 17!
Program and details: regularitylemmas.github.io/
In recent years, connections to multiaccuracy and multicalibration have led to a surge of work on regularity lemmas, with new results and applications in complexity theory, cryptography, learning theory, information theory, and communication complexity, among other areas.
We have a fantastic lineup of speakers presenting some of these recent advances! Register to join us in person or online at regularitylemmas.github.io/
(Co-organized with Parikshit Gopalan, @nhaghtal, Omer Reingold, and Salil Vadhan)
Sílvia Casacuberta retweeted
I am happy to announce Resolution's theoretical computer science (TCS) leadership!
1. Omer Reingold is joining as TCS Technical Lead, alongside his work at Stanford. ❤️
2. Tal Rabin is joining as Board Advisor, and helping with team building and leadership within TCS. ❤️
Resolution.Org Here I come!
facebook.com/omer.reingold/p…
Sílvia Casacuberta retweeted
Moritz Hardt thinks that OpenAI and Anthropic’s claims about benchmarks, which are used to test AI models’ capabilities, are misleading. But if the tests aren’t accurate, then how do we know how good the models really are? Read the interview: quantamagazine.org/he-wants-…
Sílvia Casacuberta retweeted
theorydish.blog/2026/09/29/s… blog post by theoretical computer science prof at Stanford, Omer Reingold:
“Above all, we need to reevaluate our identity. What is it that makes mathematical knowledge and research valuable? What are we offering society, and how much of it survives in a world where models match or exceed humans in some or all relevant mathematical skills? For quite some time, the implicit social contract has been that society pays us to exercise our intellectual curiosity and we, in return, provide useful skills to the next generations and practical knowledge for the world. The current crisis is driven not only by the power of LLMs, but also by how rarely we have had to examine or articulate this contract. Now that the deal needs to be renegotiated, we should approach it with humility rather than entitlement.”
Sílvia Casacuberta retweeted
Third-party auditors often audit models via the API, but do those findings transfer to chatbot interfaces?
We audited seven systems across ChatGPT, Claude, and Gemini and found that they don’t. (EMNLP 2026)
Sílvia Casacuberta retweeted
The Institute for Responsible Superintelligence (RESI): a nonprofit founded by Shafi Goldwasser, Adam Kalai, Vinod Vaikuntanathan. @ShafiGoldwasser @Vinod_MIT @AdamFungi
resi.org
Our goal: build scientific foundations needed to make superintelligence safe by design
How can theoretical computer science help us think about the interaction between algorithms and society?
I wrote some personal reflections for @KnightHennessy about my path into this area and the research community working on these questions
knight-hennessy.stanford.edu…
Sílvia Casacuberta retweeted
We have a new online boosting algorithm which is very efficient and effective. Unlike prior algorithms which maintain many weak learners and ensemble them, we operationalize the "dual view" of boosting. We don't maintain an ensemble. We try to construct an online hard core set.
Sílvia Casacuberta retweeted
On Monday @natalie_collina, Ira, and I are giving a tutorial at ICML on (multi)calibration and its applications. You can find slides and an annotated bibliography on the website: calibration-tutorial.github.… as well as an interactive demo of online calibration algs. See you there!
Sílvia Casacuberta retweeted
🔬Science can play a critical role in guiding policy interventions. How can we make our research legible to the policymakers who work on AI?
Our tutorial "From FAccT to Policy Impact" explores how we can translate our research in a specific format: the policy memo! #FAccT2026
Sílvia Casacuberta retweeted
S'il·lumina per primera vegada la torre de Jesús de la Sagrada Família 3cat.cat/3catinfo/visita-del…
In a previous work with Cynthia Dwork and Salil Vadhan (arxiv.org/abs/2312.17223, presented at STOC 2024), we showed that from multicalibration we get a stronger and more general Hardcore Lemma, from which we derive the original Hardcore Lemma with optimal density 2*delta.
...other multigroup fairness notions) has recently been shown to be powerful in other settings in TCS, including in omniprediction (arxiv.org/abs/2501.17205), in pseudoentropy characterizations (arxiv.org/abs/2507.05972), and in robust learning (arxiv.org/abs/2604.02555).
If you would like to know more, here is our recorded talk for FOCS: youtube.com/watch?v=qm3RGR07….
For an expository presentation of our results, see: ora.ox.ac.uk/objects/uuid:8c….
A similar picture emerges in the setting of complexity theory. We know, from the work of Trevisan, Tulsiani, and Vadhan, that from the Regularity Lemma (= multiaccuracy theorem) we can prove Impagliazzo’s Hardcore Lemma, albeit with suboptimal density delta.